Jan Lindström

A Distributed Real-Time Main-Memory Database for Telecommunication

Recent developments in network and switching technologies have increased the data intensity of telecommunications systems and services. The challenges facing telecom data management are now at a point that the database research community can and should become deeply involved.

Using importance of transactions and optimistic concurrency control in firm real-time databases

In a real-time database system, it is difficult to meet all of the timing constraints due to the consistency requirements of the underlying database. However, when the transactions in the system are heterogeneous, they are not all of the same importance-some are of greater importance than others. In this paper, we propose a new protocol called OCC-PDATI (Optimistic Concurrency Control Protocol using Dynamic Adjustment of serialization order and Transaction Importance), which uses information about the importance of the transactions in the conflict resolution. Performance studies of our protocol have been carried out in a prototype real-time database system. The results clearly indicate that OCC-PDATI meets the goal of favoring transactions of high importance.

Extensions to optimistic concurrency control with time intervals

Although an optimistic approach has been shown to be better suited than locking protocols for real time database systems (RTDBS), it has the problems of unnecessary restarts and heavy restart overhead. The article identifies the unnecessary restart problem in OCC-TI (Optimistic Concurrency Control with Time Intervals), proposes a solution to this problem and demonstrates that the solution will produce a correct result. Additionally, two extensions to the basic dynamic adjustment of the serialization order conflict resolution method used in OCC-TI are proposed. Experiments with a prototype implementation of a real time database system show that the proposed method clearly outperforms the original OCC-TI.Although an optimistic approach has been shown to be better suited than locking protocols for real time database systems (RTDBS), it has the problems of unnecessary restarts and heavy restart overhead. The article identifies the unnecessary restart problem in OCC-TI (Optimistic Concurrency Control with Time Intervals), proposes a solution to this problem and demonstrates that the solution will produce a correct result. Additionally, two extensions to the basic dynamic adjustment of the serialization order conflict resolution method used in OCC-TI are proposed. Experiments with a prototype implementation of a real time database system show that the proposed method clearly outperforms the original OCC-TI.

Benchmark for Real-Time Database Systems for Telecommunications

As long as there have been databases there hus been a large interest in measuring their performance. Therefore, several different benchmarks have been proposed. However, previous proposals do not consider timely response and other telecommunication application requirements. This paper shortly reviews telecommunication requirements and previous work on real-time databases. These requirements are used to prepare tile benchmark proposal for real-time database systems in telecommunication. The benchmark models a hypothetical telecommunication operator, which provides some services to subscribers. The services provided are selected to represent a wide range of real services from different application needs e.g. Intelligent Networks, 800 service and GSM user roaming.

Relaxed Correctness for Firm Real-Time Databases

Real-time database system must meet time constraints in addition to the integrity constraints. Concurrency control is one of the main issues in the studies of real-time database systems. Traditional concurrency control methods use serializability as the correctness criterion when transactions are executed concurrently. However, strict serializability as the correctness criterion is not always suitable in real-time databases. Instead, correctness requirements vary from one type of transactions to another and from one data type to another. In this paper we propose a concurrency control method witch is based on optimistic methods with a extension to relaxed serializability and semantic conflict resolution so that a special purpose concurrency control scheme can be applied. Proposed method is evaluated and tested in prototype implementation of real-time database system for telecommunications

Real Time Database Systems

A real-time database system (RTDBS) is a database system designed to handle workloads whose state is constantly changing. This system differs from traditional databases containing persistent data, mostly unaffected by time. A real-time database is a database in which transactions have deadlines or timing constraints. Real-time databases are commonly used in real-time computing applications that require timely access to data. In this article, we will discuss the most important concepts of real-time database systems. n n nKeywords: n nreal-time database system; ntransaction processing; nconcurrency control

Performance of distributed optimistic concurrency control in real-time databases

Concurrency control is one of the main issues in the studies of real-time database systems. In this paper different distributed concurrency control methods are studied and evaluated in real-time system environment. Because optimistic concurrency control is promising candidate for real-time database systems, distributed optimistic concurrency control methods are discussed more detailed way. We propose a new distributed optimistic concurrency control method, demonstrate that proposed method produces a correct results and proposed method is evaluated and tested in prototype implementation of real-time database system for telecommunications. ...

Impact of Operating System on Real-Time Main-Memory Database System’s Performance

As long as there have been databases there has been a large interest to measure their performance. However, operating system impact on database performance has not been widely studied. Therefore, this paper presents experimental results on operating system impact on database performance. Two different operating systems are studied: Linux and Chorus. Linux operating system is tested with different kernel versions and different network speeds. Chorus is used as reference point because it is a real-time operating system. Our results clearly indicate that Linux can be used as a platform for real-time main-memory databases, but the newest kernel version 2.4 should be used. Our simple experiment also confirms that the UDP gives better response time than TCP. The work done in the Linux community to reduce the long latency in the kernel has been successful and with sufficiently long request deadlines it can be used as a platform for real-time databases.